02 Selection

Figures

• Columns of data

head(malawi_MFL)

## # A tibble: 6 x 11
##   CODE  NAME  `COMMON NAME` OWNERSHIP TYPE  STATUS ZONE  DISTRICT `DATE OPENED`
##   <chr> <chr> <chr>         <chr>     <chr> <chr>  <chr> <chr>    <chr>        
## 1 MC01… A + … A+A           Private   Clin… Funct… Cent… Mchinji  Jan 1st 75   
## 2 BT24… A-C … A.C Opticals  Private   Clin… Funct… Sout… Blantyre Jan 1st 75   
## 3 MZ16… A-C … A-C Opticals  Mission/… Clin… Non-f… Nort… Mzimba   Jan 1st 75   
## 4 BT24… Akwe… Akwezeke Pvt  Private   Clin… Funct… Sout… Blantyre Jan 1st 75   
## 5 BT24… AB M… Abowa         Private   Clin… Funct… Sout… Blantyre Jan 1st 75   
## 6 LL04… ABC … ABC Clinic    Christia… Hosp… Funct… Cent… Lilongwe Jan 1st 75   
## # … with 2 more variables: LATITUDE <chr>, LONGITUDE <chr>

head(malawi_WHO)

## # A tibble: 6 x 10
##   Country Admin1 `Facility name` `Facility type` Ownership   Lat  Long
##   <chr>   <chr>  <chr>           <chr>           <chr>     <dbl> <dbl>
## 1 Malawi  Centr… 80 Block Clinic Clinic          MoH       -12.9  33.4
## 2 Malawi  Centr… ABC Community … Clinic          FBO       -14.0  33.7
## 3 Malawi  Centr… Adventist Heal… Health Centre   FBO       -14.0  33.8
## 4 Malawi  Centr… Alinafe Commun… Community Hosp… FBO       -13.4  34.2
## 5 Malawi  Centr… Area 18 Health… Health Centre   MoH       -13.9  33.8
## 6 Malawi  Centr… Area 25 Health… Health Centre   MoH       -13.9  33.8
## # … with 3 more variables: `LL source` <chr>, iso3c <chr>,
## #   facility_type_9 <chr>

• Facility plots

• Maps

Missing values/wrong coordinates

• 181 in the MFL (this includes NA’s and coordinates that did not intersect with admin levels) and 9 in the WHO data. None are the same facilities/share same names.

# MFL - 1546 before omitting NA's, then 1427.
# 62 coordinates did not intersect with admin data - now 1365

# using anti_join() to see which facilities were omitted 
df_malawi_facilities_MFL = as.data.frame(malawi_facilities_MFL) # convert to data frame to use in anti_join

NA_MFL = anti_join(malawi_MFL, df_malawi_facilities_MFL) # 119 missing coordinates in MFL

# adding the number of wrong coordinates - 61
intersect_admin1_MFL = st_intersection(malawi_facilities_MFL, malawi_admin1) # returns data frame of matches

intersect_admin1_MFL = as.data.frame(intersect_admin1_MFL) # convert to data frame to use in anti_join

no_intersect_admin1_MFL = anti_join(malawi_facilities_MFL, intersect_admin1_MFL) # working out which facilities did not intersect

no_intersect_admin1_MFL_1 = no_intersect_admin1_MFL[ ,-c(10, 11)]
NA_MFL_1 = NA_MFL[, -c(10, 11)]
no_intersect_admin1_MFL_1 = st_set_geometry(no_intersect_admin1_MFL_1, NULL)
names(NA_MFL_1)[3] = "COMMON.NAME"
names(NA_MFL_1)[9] = "DATE.OPENED"

NA_MFL = rbind(NA_MFL_1, no_intersect_admin1_MFL_1) # combined dataframe of NA and wrong coordinates in MFL

# WHO - 648 with 9 missing coordinates, now 639
NA_WHO = anti_join(malawi_WHO, new_malawi_WHO)

Proportion that are privately owned in the MFL data

• 30.3% of facilities are privately owned (433/1427)
  
• Top 3 types are clinics (356), dispensaries (45) and hospitals (16)

private_MFL = filter(malawi_facilities_MFL, OWNERSHIP == "Private")
# 433 private facilities, 30.3%

# which types are private?
private_MFL$TYPE = as.factor(private_MFL$TYPE)
types_private_MFL = as.data.frame(table(private_MFL$TYPE))
types_private_MFL

##            Var1 Freq
## 1        Clinic  356
## 2    Dispensary   45
## 3 Health Centre    9
## 4   Health Post    1
## 5      Hospital   16
## 6       Private    6

Proportion of facility types in each district

# district per facility (does not include the 62 facilities) MFL
intersect_admin2_MFL = st_intersection(malawi_facilities_MFL, malawi_admin2)
intersect_admin2_MFL$TYPE = as.factor(intersect_admin2_MFL$TYPE)
intersect_admin2_MFL$shapeName = as.factor(intersect_admin2_MFL$shapeName)

# types by district
by_district_MFL = as.data.frame(table(intersect_admin2_MFL$TYPE, intersect_admin2_MFL$shapeName))
head(by_district_MFL)

##                Var1   Var2 Freq
## 1  Central Hospital Balaka    0
## 2            Clinic Balaka   10
## 3        Dispensary Balaka    1
## 4 District Hospital Balaka    1
## 5     Health Centre Balaka   11
## 6       Health Post Balaka    4

# WHO 
intersect_admin2_WHO = st_intersection(sf_malawi_WHO, malawi_admin2)
intersect_admin2_WHO$Facility.type = as.factor(intersect_admin2_WHO$Facility.type)
intersect_admin2_WHO$shapeName = as.factor(intersect_admin2_WHO$shapeName)

by_district_WHO = as.data.frame(table(intersect_admin2_WHO$Facility.type, intersect_admin2_WHO$shapeName))
head(by_district_WHO)

##                 Var1   Var2 Freq
## 1   Central Hospital Balaka    0
## 2  District Hospital Balaka    1
## 3   Mission Hospital Balaka    0
## 4     Rural Hospital Balaka    0
## 5 Community Hospital Balaka    0
## 6      Health Centre Balaka   13

# Example pie charts for district 

## Balaka 
balaka_MFL = filter(by_district_MFL, Var2 == "Balaka")
balaka_MFL = filter(balaka_MFL, Freq > 0)
balaka_MFL = balaka_MFL[ , -c(2)]

## plot
balaka_plot_MFL = sunburst(balaka_MFL, count = TRUE, legend = list(w=120))
balaka_plot_MFL

Legend

## 
balaka_WHO = filter(by_district_WHO, Var2 == "Balaka")
balaka_WHO = filter(balaka_WHO, Freq > 0)
balaka_WHO = balaka_WHO[ , -c(2)]

## plot
balaka_plot_WHO = sund2b(balaka_WHO, rootLabel = "Facilities")
balaka_plot_WHO

## Lilongwe 
lilongwe_MFL = filter(by_district_MFL, Var2 == "Lilongwe")
lilongwe_MFL = filter(lilongwe_MFL, Freq > 0)
lilongwe_MFL = lilongwe_MFL[ , -c(2)]

## plot
lilongwe_plot_MFL = sund2b(lilongwe_MFL, rootLabel = "Facilities")
lilongwe_plot_MFL

## 
lilongwe_WHO = filter(by_district_WHO, Var2 == "Lilongwe")
lilongwe_WHO = filter(lilongwe_WHO, Freq > 0)
lilongwe_WHO = lilongwe_WHO[ , -c(2)]

## plot
lilongwe_plot_WHO = sund2b(lilongwe_WHO, rootLabel = "Facilities")
lilongwe_plot_WHO

Duplicates within dataset

• 22 names in MFL appear more than once, of these Mpepa Health Centre (2 decimal places), Liwonde Medical Clinic (identical), Ngatala Health Post (3 decimal places), Cape Maclear (identical) and LIFE STYLE PVT CLINIC (identical) have similar/same coordinates

• Of these, Ngatala, Cape Maclear and LIFE STYLE PVT CLINIC are most likely the same entries as other attributes are also the same. The other facilities differ in type and so might not be duplicates.

• 10 names in WHO appear more than once, Mkango Health Centre has same coordinates (4 decimal places) and other attributes are also the same

• Duplicate coordinates: in the MFL, 44 coordinates have been repeated at least more than once.
  Interesting note is that 24 facilities in Blantyre have reused the same coordinates (34.3015278, -13.2512161) and they consist of 1 hospital and a mix of clinics and health centres. All have unique names as well.
  Same coordinates (33.7895208991213, -14.0017665100358) have also been used for 4 facilities, all in different districts.
  Some results are due to no coordinates being available and (-1,1) for example is inputted instead.
  No duplicates in the WHO data

• Duplicates between datasets?

# MFL
duplicate_name_MFL = malawi_facilities_MFL[malawi_facilities_MFL$NAME %in% malawi_facilities_MFL$NAME[duplicated(malawi_facilities_MFL$NAME)], ]

# WHO 
duplicate_name_WHO = new_malawi_WHO[new_malawi_WHO$`Facility name` %in% new_malawi_WHO$`Facility name`[duplicated(new_malawi_WHO$`Facility name`)], ]

# Same coordinates within dataset

## MFL 
duplicate_coord_MFL = malawi_facilities_MFL[malawi_facilities_MFL$geometry %in% malawi_facilities_MFL$geometry[duplicated(malawi_facilities_MFL$geometry)], ]

## WHO 
sf_malawi_WHO$geometry[duplicated(sf_malawi_WHO$geometry)] # none 

## Geometry set for 0 features 
## bbox:           xmin: NA ymin: NA xmax: NA ymax: NA
## geographic CRS: WGS 84

# Blantyre 
blantyre_duplicate_coord_MFL = filter(duplicate_coord_MFL, DISTRICT == "Blantyre")
blantyre_MFL = filter(by_district_MFL, Var2 == "Blantyre")
blantyre_MFL = filter(blantyre_MFL, Freq > 0)
blantyre_MFL = blantyre_MFL[ , -c(2)]

## plot
blantyre_plot_MFL = sund2b(blantyre_MFL, rootLabel = "Facilities")
blantyre_plot_MFL

Facilities within 50km